• Articles
• Press Releases
• Events

To learn how the eMeter solution can increase efficiency, reduce risk, and improve operations for your business, call us at (650) 227-7770 or email sales@emeter.com.

Articles

Integrating Residential Dynamic Pricing and Load Control: The Literature

Energy Pulse, December 14, 2005 by Chris King

Since electricity cannot be stored, power utilities face the problem of balancing supply and demand in real time. In addition to using supply side resources – peaker plants – utilities also use demand reduction programs. The programs include dynamic pricing programs and load control programs. This paper reviews the publicly-available literature regarding the interaction of dynamic pricing and automated control programs and finds that these programs are complementary; that is, an integrated program that combines pricing and automated control typically results in peak demand reductions that are up to twice as large as the reduction from either a dynamic pricing or load control program alone.

Program History

Residential dynamic pricing programs include hourly pricing and critical peak pricing programs. Only one residential hourly pricing program has been described in the literature, that of Commonwealth Edison of Chicago, which began in January 2003.1 In Commonwealth’s program, customers are told a day in advance of hourly prices and are given special notice when prices exceed a critical threshold, not expected to exceed more than a dozen or so times per year. At least five critical peak pricing programs have been implemented, including California’s Statewide Pricing Pilot, which began in July 2003. Other critical peak pricing programs with published results are being or have been operated in Pennsylvania, Florida, Texas, Ohio, Indiana, Michigan, Virginia, and France. Prices in these critical peak pricing programs have all included daily peak and off-peak time-of-use pricing, with critical peaks dispatched up to 15 times per year.

Of the nine residential dynamic pricing programs described in the literature, three have been pricing only, and seven have integrated pricing with automated response. These seven integrated programs were or are in California (two), Pennsylvania, Georgia, Florida, Texas, and a multi-state program including Ohio, Indiana, Michigan, and Virginia. All seven integrated programs had or have critical peak pricing with automated control in which air conditioner thermostats are automatically adjusted when critical peak prices are in effect. One program, California’s Advanced Demand Response Systems (ADRS) pilot (a part of the Statewide Pricing Pilot), includes control of other appliances as well.

Load control has been used by many utilities in the U.S. and internationally to reduce electricity demands during critical peak periods. Over 450 U.S. utilities have tested or implemented such programs, according to the Electric Power Research Institute.2 In these programs, the utility issues a signal at times of system peak to turn off customers’ air conditioners and, in some cases, electric water heaters and swimming pool pumps. Electric space heaters are controlled at some winter-peaking utilities. Dispatch is usually limited to no more than 12 to 15 days per year. Customers typically receive a bill discount of a few dollars per month during the summer months as an incentive to participate.

In integrated dynamic pricing and load control programs, customers do not receive a monthly incentive payment. Instead, their bill savings are calculated based on the actual amount of load reduction during peak hours, as recorded by an advanced meter installed at their residence. The savings amount depends on the dynamic pricing rate in effect.

Program Results

Program results for dynamic pricing and load control programs are reported in various ways, usually as the effect of the program in reducing peak demand, the goal of most programs. This effect is expressed as a percentage of peak demand or in kilowatts per customer. Dynamic pricing program results usually are the form of percentage peak demand reductions, but often include customer price elasticity, which reflects how much demand reduction results from a given change in peak or critical peak price. Load control program results are presented in kilowatts per customer of peak reduction. Since incentive payments for load controls are not directly linked to individual customer behavior, elasticities cannot be estimated for load control programs.

Program results for dynamic pricing programs are measured via advanced metering of participants, with load reductions usually estimated by comparing participants to a control group. Results for load control programs are usually based on engineering estimates using population appliance data, number of load control devices installed, and load research data, including temperature data. Some load control programs have reported results using advanced metering data from a sample of program participants.

A broad search of the publicly-available literature resulted in 24 programs with sufficient data to include in the sample. Sixteen detailed papers were available for load control programs, from the over 450 programs that have been conducted. One reason few published papers are available is that most utilities rely on engineering estimates rather than conducting metered load studies.3 The program results from the literature were converted to a consistent basis of percentage reduction in peak demand on critical peak days for each program. The percentages are the peak demand reduction for participating customers only. Table 1 lists the programs included; Figure 1 summarizes the comparison results.

The load reductions for integrated dynamic pricing and automated control programs were, on average, 53 percent larger than load reductions for load control programs alone and 102 percent larger than load reductions for dynamic pricing programs alone. The load reductions for load control programs were, on average, 32 percent higher than those for dynamic pricing programs. However, load control program participants have peak loads that are, on average, 82 percent higher than peak loads for dynamic pricing program participants, reflecting that load control programs are limited to households with central air conditioning.

Table 1 – Residential Load Control and Dynamic Pricing Programs

Figure 1 – Comparison of Pricing Only, Load Control Only, and Integrated Programs(4)

The load control program results varied more widely than the other program results, reflecting greater diversity within the program category. Residential dynamic pricing program load reductions ranged from 18 to 27 percent. Combined pricing and automated control program load reductions ranged from 38 to 53 percent. In contrast, direct load control program load reductions ranged from 16 percent to 61 percent.

The wider range of reported reductions for load control programs reflect three factors. First, load control program operations range from 50 percent cycling (in which air conditioners are turned off 30 minutes of every hour) to 100 percent shedding (in which air conditioners are turned off continuously for the critical peak period). Second, load control results are estimated from the number of control points installed and weather data, rather than being measured via metered data for individual customers. Third, a source of variation in direct control programs utilizing one-way communications is control switch operating reliability. The Electric Power Research Institute found that most load control programs utilize one-way communications and do not include verification of demand reduction at the customer level. (5) Two-way communications obviates this issue.

Discussion

The reported peak reductions of the different program types strongly suggest that pricing and load control strategies have synergistic effects. The integrated load control and dynamic pricing programs provided an average 53 percent more load reduction than load control alone and an average 104 percent greater load reduction than dynamic pricing alone. There are three important reasons for this synergy.

First, dynamic pricing provides an economic incentive for load reduction to all household appliances while load control in the reported programs is limited in summer to air conditioning, electric water heating, and pool pumps. Figure 3 shows the annual consumption by residential appliance for California, and Figure 4, more importantly, shows the residential air conditioning load compared to all other residential load at the time of the system peak.

Figure 3 – California Residential Electricity Annual Consumption by End Use (6)

Figure 4 – California Residential End Use Coincident Peak Demand (7)

The results of dynamic pricing programs shows that applying economic incentives for load reduction beyond air conditioning results in customers reducing load beyond air conditioning.

Second, dynamic pricing closes the customer feedback loop, providing more direct and individualized economic incentives than load control incentives. With dynamic pricing integrated into a control program, incentives are based on the magnitude of individual action: more load reduction means more savings. Without integrated pricing, load control programs give the customer not an incentive to reduce load, but an incentive to reduce discomfort. One result is that the customers most likely to remain in a load control program are those with oversized air conditioning units, an important factor when over 20% of participants can be expected to drop out over time. (8) Another discomfort avoidance would be to install insulation, which is desirable for energy efficiency and ongoing demand reduction, but decreases the load reductions associated with the direct control program. Another example is customers not reporting load control switch failures (customers are not usually aware of them), and, in the extreme, disabling load control switches. In all these cases, the customer continues to receive the load control incentive payment, regardless of whether load control is ineffective as a result of oversized air conditioners, insulation, switch failures, disabled switches, or failure of communications to reach the switch.

The public literature provides limited information regarding the number of oversized (or turned off) air conditioners and switch failures. Two studies quantified the number of air conditioners not in use and load control switch failures (Table 2). In addition, San Diego Gas & Electric cited “unreliable switches” as one of several factors leading to the termination of its residential load control program in 1993; however, no data were provided. (9)

Table 2 – Residential Load Control Program Reliability Results

Third, automated control makes it much easier for customers to respond to dynamic price signals. Automation reduces or eliminates the need to “run around turning off all the lights.” Residential customers have shown neither the desire nor the ability to monitor constantly changing prices, making automation a highly desirable means of reducing load. As a result, customers in integrated dynamic pricing and automated control programs report high levels of satisfaction with the automated control capability. (10) An important feature of these programs is that customers decide at which price levels load should be controlled and by how much. For example, a customer may choose to raise his thermostat to 80 degrees during peak hours and to 85 degrees during critical peak hours.

Conclusion

The literature regarding the interaction of dynamic pricing and automated control program strongly suggests that these programs are complementary; that is, a program that combines dynamic pricing and automated control typically results in peak demand reductions much larger than either a dynamic pricing or load control program alone. The magnitude of the increased load reductions ranges from 53 percent to 104 percent (i.e. 53 percent to 104 percent more “negawatts” are provided). There are three reasons for the greater response. First, more load is subjected to economic incentives, in all about twice as much as with air conditioning control only. Second, dynamic pricing closes the feedback loop, thus providing a much stronger economic incentive than once-a-month payments typical of load control only programs. Third, automated response to dynamic pricing makes it much easier for consumers to respond to price signals, thus increasing both demand reductions and customer satisfaction.

Bibliography

• Aubin, Christophe et al. "Real-Time Pricing of Electricity for Residential Customers: Econometric Analysis of an Experiment.” Journal of Applied Econometrics. Dec. 1995.
• Baxter, T. et al. “An Application of End-Use Load Data Analysis: Estimating Cycling Switch Effectiveness and Freeridership in a Residential A/C Cycling Program.” Proceedings EPRI Conference Load Management: Dynamic DSM Options for the Future, May 1994.
• Baxter, T. et al. “Field Measurement of Load Availability and Load Control Switch Effectiveness in a Residential AC Cycling Pilot.” ACEEE Summer Study Proceedings 1994.
• Barakat & Chamberlin. "Impact of Demand-Side Management on Future Customer Electricity Demand: An Update.” Final Report for Edison Electric Institute and Electric Power Research Institute. September 1990.
• Braithwait, S.D. "Residential TOU Price Response in the Presence of Interactive Communication Equipment.” In Pricing in Competitive Electricity Markets, Ch 20, Kluwer Academic Publishers, 2000.
• Budd, C., “Making Electricity Markets Work: Hourly Prices for the Home,” EUCI Load Management Conference, October 2003.
• C Three Group, “Long Island Power Authority Profile,” in Final Report Load Management 2001, October 2001.
• California Energy Commission, “1998 Baseline Energy Outlook,” August 1998.
• Electric Power Research Institute, “1992 Survey of Utility Demand-Side Management Programs,” Report TR-102193, May 1993.
• Energy Information Administration, “1997 Electric Air-Conditioning Consumption Tables,” November 1999.
• Goldberg, M. “Knowing Your Limits: Direct Load Control Capacity Credits Based on Censoring Distribution Analysis,” AEIC Load Research Conference, August 2000.
• Hackner, R. “The Evaluation of Residential Air Conditioner Direct Load Control Programs at Five Midwestern Utilities,” Proceedings EPRI Conference Load Management: Dynamic DSM Options for the Future, May 1994.
• Hartway, Rob et al. “Smart meter, customer choice and profitable time-of-use rate option.” Energy – The International Journal. Vol. 24, pp. 895-903. October 1999.
• Levy, R. “TranstexT Advanced Energy Management System,” July 1994.
• Levy, R. et al. “Using Demand Response Management to Achieve Utility Load Shape Objectives,” EPRI Report EP-P6035/C3047, March 2002.
• Moe, M. “Eliminating ‘Free Riders’ in A/C Cycling Programs,” Proceedings EPRI Conference Load Management: Dynamic DSM Options for the Future, May 1994.
• Pacific Gas & Electric et al. “Monthly Report on Statewide Pricing Pilot to California Public Utilities Commission and California Energy Commission,” December 15, 2003.
• Parker, D. S. "Research Highlights from a Large Scale Residential Monitoring Study in a Hot Climate." Proceedings of International Symposium on Highly Efficient Use of Energy and Reduction of its Environmental Impact, pp. 108-116, Japan Society for the Promotion of Science Research for the Future Program, JPS-RFTF97P01002, Osaka, Japan, January 2002.
• Peak Load Management Alliance. “Demand Response Awards for the Year 2001.”
• Results Center. “Sacramento Municipal Utility District Residential Peak Corps,” 1994.
• Rocky Mountain Institute. “ADRS Load Impact Interim Results,” Presentation to CPUC Working Group 3, R.02-06-001, November 8, 2004.
• Riordan, K. “Assumptions to Actuals: Progress of A/C Load Control Program,” Proceedings of Fourth International Symposium on Distribution Automation and Demand Side Management, January 1994.
• Southern California Edison, “Phase 2 of 2003 General Rate Case Load Control Proposal,” Before the Public Utilities Commission of the State of California, March 2003.
• Spellman, R. “Utility Experience with Load Management in Texas,” Proceedings EPRI Conference Load Management: Dynamic DSM Options for the Future, May 1994.
• Zeanah, J. and S. Kammerer. “Customer Response to the TranstexT AEM System,” Final Report to Gulf Power Company, April 1994.

Endnotes

1. Budd, C., “Making Electricity Markets Work: Hourly Prices for the Home,” EUCI Load Management Conference, October 2003.
2. Electric Power Research Institute, “1992 Survey of Utility Demand-Side Management Programs,” Report TR-102193, May 1993.
3. Riordan, K., “Assumptions to Actuals: Progress of A/C Load Control Program,” Proceedings of Fourth International Symposium on Distribution Automation and Demand Side Management, January 1994.
4. See table
5. Op. cit.
6. California Energy Commission, “1998 Baseline Energy Outlook,” August 1998.
7. Op. cit.
8. Results Center, “Sacramento Municipal Utility District Residential Peak Corps,” 1994.
9. SDG&E Advice Letter 1320-E to the California Public Utilities Commission, May 1, 2001.
10. Levy, R. et al. “Using Demand Response Management to Achieve Utility Load Shape Objectives,” EPRI Report EP-P6035/C3047, March 2002